Now able to run as a standalone module or shell program, with limited functionality #146
Conversation
| # TODO: go through subdirectories if directory/ies specified | ||
| all_filenames = args.files |
There was a problem hiding this comment.
🤷♂️, I'm pretty happy to say that you have to pass exact files if you're passing files.
There was a problem hiding this comment.
I figured, since that's how shed does it, but I would like to support as much of flake8's functionality as possible to ease any transition from running [name of this project] as a flake8 plugin to running it standalone. Ofc flake8 is kinda crazy so I won't bother supporting everything (e.g. --extend-ignore, --ignore, --select, --extend-select I think kinda just makes everything more complicated), but I think supporting directories shouldn't be very complicated and doesn't have much downsides.
| for file in all_filenames: | ||
| plugin = Plugin.from_filename(file) | ||
| for error in sorted(plugin.run()): | ||
| print(f"{file}:{error}") |
There was a problem hiding this comment.
A small ProcessPoolExecutor invocation here can make it n_cores times faster for large repos... although we'd still want to sort by file and lineno. Let's do that later.
First steps towards having it run as a standalone program, which is required for autofixing.
_get_git_repo_root.flake8_trio/__main__to run as a module, and aconsole_scriptsentry pointtest_flake8_trio, and added a couple tests to test the aboveAlso:
flake8-bandit- runninggitdirectly is .. kinda sus https://flake8.codes/S607/ ; but ifsheddoes it I'm not gonna bother, and I'm very much just disabling ~all errors bandit throws, so it's not really contributing anything.reportUnusedCallResultis frustrating, byebye~Remaining steps:
--enable-visitor-codes-regexadded these to OP in Switch backend to libCST #124